Span Semantic Role Labeling on CoNLL-2012 (OntoNotes) v5.0 (test)
88.3F1 ScoreZhang et al.
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Zhang et al.Syntax Category=syntax-agnostic, Backbone=RoBERTa2021.09 | 88.3 | 88.1 | 88.6 | |
| OursSyntax Category=syntax-agnostic, Backbone=RoBERTa2021.09 | 88.3 | 88.6 | 87.9 | |
| OursSyntax Category=syntax-agnostic, Backbone=BERT2021.09 | 87.8 | 88 | 87.7 | |
| Paolini et al.Syntax Category=syntax-agnostic, Backbone=T52021.09 | 87.7 | — | — | |
| Conia and NavigliSyntax Category=syntax-agnostic, Backbone=BERT2021.09 | 87.3 | 86.9 | 87.7 | |
| Blloshmi et al.Syntax Category=syntax-agnostic, Backbone=BART2021.09 | 87.3 | 87.8 | 86.8 | |
| Marcheggiani and TitovSyntax Category=syntax-aware, Backbone=RoBERTa2021.09 | 86.8 | 86.5 | 87.1 | |
| Jindal et al.Syntax Category=syntax-agnostic, Backbone=BERT2021.09 | 86.6 | 86.3 | 86.8 | |
| Shi and LinSyntax Category=syntax-agnostic, Backbone=BERT2021.09 | 86.5 | 85.9 | 87 | |
| UniSRLSystem Type=Single2019.01 | 86 | 85.7 | 86.3 | |
| Li et al.Syntax Category=syntax-agnostic2021.09 | 86 | 85.7 | 86.3 | |
| He et al. (2018a)System Type=Single2019.01 | 85.5 | — | — | |
| Peters et al. (2018)System Type=Single2019.01 | 84.6 | — | — | |
| Tan et al. (2018)System Type=Ensemble2019.01 | 83.9 | 83.3 | 84.5 | |
| He et al. (2017)System Type=Ensemble2019.01 | 83.4 | 83.5 | 83.3 | |
| Tan et al. (2018)System Type=Single2019.01 | 82.7 | 81.9 | 83.6 | |
| He et al. (2017)System Type=Single2019.01 | 81.7 | 81.7 | 81.6 | |
| Zhou and Xu (2015)System Type=Single2019.01 | 81.3 | — | — | |
| FitzGerald et al. (2015)System Type=Ensemble2019.01 | 80.1 | 81.2 | 79 | |
| Pradhan et al. (2013) (Revised)System Type=Single2019.01 | 77.5 | 78.5 | 76.6 | |
| UniSRLEvaluation Setup=End-to-end, Model Variant=Single2019.01 | 0.831 | 0.849 | 0.814 | |
| He et al. (2018a)Evaluation Setup=End-to-end2019.01 | 0.829 | 0.819 | 0.84 | |
| Strubell et al. (2018)Evaluation Setup=End-to-end2019.01 | 0.799 | 0.807 | 0.791 | |
| He et al. (2017)Evaluation Setup=End-to-end, Model Variant=Ensemble2019.01 | 0.784 | 0.802 | 0.766 | |
| He et al. (2017)Evaluation Setup=End-to-end, Model Variant=Single2019.01 | 0.768 | 0.786 | 0.751 |